intersection delay modeller - emme modeller naveen juvva aecom 21st international emme users'...

Intersection Delay Modeller - EMME Modeller

Naveen JuvvaAECOM

21st International EMME Users' Conference

2011 International EMME Users' Conference 2

Outline

• Background• Why the Modeller Tool?• Intersection Delays in EMME• Intersection Delay Methodology• Intersection Delay Calibration• Construct of the Modeller Tool• Travel Time Contour Results• Observations and Conclusions09/15/2011


Background

• Travel time contours from the Regional model not realistically depicting existing conditions– Incorporating turn delays

• A post-assignment calculation of intersection turn delays

• Application of the intersection turn delays to feed back into the route choice

• Desired: Tool to run the above procedures09/15/2011


Why the Modeller Tool

• An independent utility that can be run in any setting• Flexibility in terms of talking various components of

EMME desktop and prompt console• Modeller still provides connectivity to the macro

platform, which can be called up if needed• Python programming provides better organization of

code compared to steps in macros• A better troubleshooting support in the form of

logbook

09/15/2011


Intersection Delays in EMME

• Previous implementation of intersection delays in EMME– HCM Based turn volume delay functions: Terry

Patridge– Intersection delay functions: Aashtiani and Iravani

• Options:– As a postprocessing tool to create network

analysis– As part of the route choice

09/15/2011


Intersection Delay Methodology

• Intersection control is input where available; elsewhere, a rule-based intersection control is applied

• Signals: Cycle length and splits are based on:– High – High

– High – Low

– Low – Low

• Unsignalized– Two-way stop-controlled

– All-Way stop-controlled

• HCM Delay Equations with some simplified assumptions09/15/2011


Intersection Delay Calibration

• Calibration of the intersection delay methodology– Corridor travel times: opportunity to calibrate

signalized and unsignalized intersections– Resulting intersection turn delays and travel route

patterns confirmed observations from the field

09/15/2011



09/15/2011

Highw

ay 403

407 ETR

GO Transit Station

Trafalgar

GO Transit Station

407 ETR

Trafalgar

Shortest path with only link travel times Shortest path with link travel times and intersection delays



• Calibration of the intersection delay methodology• Sample corridor travel time comparison

• Directionality was a problem with the model – independent of the intersection delays

09/15/2011

Field travel times (min)

With only link delays

(min)

With link and turn delays

(min)

Southbound 14.6 7.8 13.9

Northbound 13.8 8.8 14.8


Construct of the Tool

• Import the python modules– eg: from __future__ import with_statement

• Define namespaces– eg: NETWORK_TOOL_NAMESPACE = "inro.emme.standard.data.network."

• Pagebuilder, etc• Logbook trace• Network, matrix and function editing procedures• Run macros within to execute sub procedures• Use desktop API reference: to access worksheets, etc

09/15/2011


Construct of the Toolwith _m.logbook_trace( name="1.2 Calculate turn delays for high-high signals"): spec = { "type": "NETWORK_CALCULATION", "result": "@ptime", "expression": "(1/60)*(.5 * 120 * put(1 - put(15 / 120)) * get(2) / (1 - (1 .min. put( @pvol / 1700 * get(1))) * get(1)) \ + 900 * 1 * ((put(get(3) - 1))+ sqrt(get(4) * get(4) + 8 * .5 * get(3) / (1700 * get(1) * 1))))*(-105.le.@angle)*(@angle.le.-75)\ +(1/60)*(.5 * 120 * put(1 - put(60 / 120)) * get(2) / (1 - (1 .min. put(@pvol / 1900 * get(1))) * get(1))\ + 900 * 1 * ((put(get(3) - 1))+ sqrt(get(4) * get(4) + 8 * .5 * get(3) / (1900 * get(1) * 1))))*(-15.le.@angle)*(@angle.le.15)\ +.08*(75.le.@angle)*(@angle.le.105)", "selections": { "incoming_link": "@intj=1,2 and ul3=9999 or ul3=800,99999", "outgoing_link": "@int=1,2 and ul3=800,99999" } } compute_network(spec, scenario)

09/15/2011


Implementation of the Tool

09/15/2011

Custom Tool

Tool Run status


Implementation of the Tool

09/15/2011


Travel Time Contours

09/15/2011

With only link travel times considered



09/15/2011

With link travel times and turn delays considered



09/15/2011

With link travel times and turn delays consideredTurn delays fed into route choice


Findings from Intersection Modeller

09/15/2011

• Intersection turn delays reflect fairly closely the delays at signalized intersections– Protected left turns were not explicitly modelled

• Delays for minor approaches at TWSC and all approaches at AWSC are unreasonably high in some cases where the opposing turn volume (from oppvol.mac) are very high.– An upper limit was imposed– These are the cases where turn delays influenced route

choice when they were incorporated into turn penalty functions


Not So Sweeping Conclusions• Trouble shooting using logbook

– The organized nature of the logbook trace is helpful to detect errors

• Run macros as part of the tool

– Easier to call macros (at least for some users) in some cases

• Better organization and flexibility of code compared to macros – use of python scripting language medium

• Modeller is particularly helpful in creating utility tools that can be delivered to the client for them to incorporate into their toolbox

• Learning curve to create sophisticated tools or implement entire models in the Modeller

• Implementation of a particular travel demand model in the Modeller without use of other components of EMME

09/15/2011

intersection delay modeller - emme modeller naveen juvva aecom 21st international emme users'...

Documents